This disclosure relates signaling of adaptive resolution for motion vector difference in video encoding and decoding. An example method is disclosed decoding one or more video blocks of a video stream. The method may include receiving the video stream; extracting an inter-prediction syntax element from the video stream to determine whether an ADAPTMV mode is signaled for one or more video blocks, the ADAPTMV mode being a single-reference inter-prediction mode with adaptive motion vector difference (MVD) pixel resolution; determining a current MVD pixel resolution associated with the one or more video blocks abased on whether the ADAPTMV mode is signaled in the inter-prediction syntax element; and extracting and decoding one or more MVD-related syntax elements associated with the one or more video blocks based on whether the ADAPTMV mode is signaled in the inter-prediction syntax element and the current MVD pixel resolution.
Legal claims defining the scope of protection, as filed with the USPTO.
2. The method of claim 1, wherein the inter-prediction syntax element is signaled in the video bitstream after an inter-prediction reference frame index associated with the one or more video blocks is signaled.
4. The method of claim 3, wherein a context for decoding the ADAPTMV mode in the inter-prediction syntax element is shared with decoding of the NEWMV mode, the NEARMV mode, and the GLOBALMV mode.
5. The method of claim 1, wherein the method further comprises deriving one or more contexts for decoding one or more MVD-related syntax elements associated with the one or more video blocks based on whether the ADAPTMV mode is signaled in the inter-prediction syntax element.
7. The method of claim 6, wherein when the one or more video blocks are encoded in the ADAPTMV mode, a first context is derived for decoding the first MVD syntax element or the third MVD syntax element whereas when the one or more video blocks are coded in an inter-prediction mode other than the ADAPTMV mode, a second context distinct from the first context is derived for decoding the first MVD syntax element or the third MVD syntax element.
8. The method of claim 1, wherein the current MVD pixel resolution for the one or more video blocks is 2n-pel (2n pixels), where n is an integer and is between −6 and 11 inclusive.
9. The method of claim 8, further comprising deriving, from the video bitstream, an MVD magnitude range associated with the one or more video blocks, wherein fractional MVD pixel resolution is allowed only when the MVD magnitude is equal to or is lower than a predetermined threshold MVD magnitude.
10. The method of claim 9, wherein allowed MVD pixel resolutions correspond to different MVD magnitudes in a non-ascending order.
11. The method of claim 8, further comprising obtaining, from the video bitstream, an MVD class index for specifying an MVD magnitude range associated with the one or more video blocks.
12. The method of claim 11, wherein a fractional MVD pixel resolution is allowed only when the MVD class index equals to or is lower than a predetermined threshold MVD class index.
13. The method of claim 12, wherein MVD class indices equal to or higher than the predetermined threshold MVD class index are each associated with a single allowed integer MVD pixel resolution value.
14. The method of claim 11, wherein allowed MVD pixel resolutions correspond to different MVD class indices in a non-ascending order.
15. The method of claim 14, wherein MVD pixel resolutions associated with different MVD class indices are distinct.
17. The computing system of claim 16, wherein the inter-prediction syntax element is signaled in the video bitstream after an inter-prediction reference frame index associated with the one or more video blocks is signaled.
19. The computing system of claim 18, wherein a context for encoding the ADAPTMV mode in the inter-prediction syntax element is shared with encoding of the NEWMV mode, the NEARMV mode, and the GLOBALMV mode.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
May 25, 2022
November 26, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.